(Press-News.org) Computer scientists and weather scientists have taken the first steps toward creating an AI agent capable of analyzing and answering questions in natural language, such as English, about data from AI-driven weather and climate forecasting models.
The research team from the University of California San Diego will present the first AI weather agent they developed, named Zephyrus, at the 14th International Conference on Learning Representations (ICLR) April 23–27 in Rio de Janeiro.
Recently, models driven by AI and deep learning have considerably improved weather forecasting. But analyzing the resulting data remains difficult and time-consuming. A main issue is that these types of AI models are not able to describe their findings in plain language. A secondary issue is that these models are not able to reason about text information, such as meteorology reports and weather bulletins. The UC San Diego research team aims to address both.
“Our goal is to increase access to critical data and predictions by lowering the barrier to entry to analyzing these data,” said Duncan Watson-Parris, a study co-author and faculty member at the UC San Diego Scripps Institution of Oceanography. “We want to increase the speed with which we can reason about multimodal data and learn about the Earth by making it easier for students and young scientists to interact with different datasets.”
The researchers also hope the findings will lead to AI agents that will be able to bring similar advances to other disciplines, especially climate science. Meteorology was a perfect test case because it combines large, complex datasets that change over time and the need to reason about these data in plain language. “Weather prediction is a critical scientific challenge, with profound implications spanning agriculture, disaster preparedness, transportation, and energy management,” the researchers write.
To bridge the gap between a code-driven AI model and language-based AI agent, the researchers set up an environment that allows the agents to interact with weather models and data via code. The AI agent is capable of handling language-based queries, translating them into code, and then translating the code-generated answers into plain language.
Zephyrus performed well on simple tasks, such as finding locations with specific weather conditions, as well as weather forecasts for specific locations at certain times. But it struggles with finding locations with extreme weather and report generation. Researchers tested four frontier LLMs to power Zephyrus, and all performed with similar accuracy.
For the next iteration of the AI agent, researchers plan to use larger training datasets. Next steps also include fine-tuning open-source models for climate-focused tasks.
“Our vision is to democratize earth science. Zephyrus is a crucial step toward creating AI co-scientists that dramatically lower the barrier to entry, allowing students and researchers everywhere to access and reason about critical weather and climate data at unprecedented speeds,” said Rose Yu, study co-author and a faculty member in the UC San Diego Department of Computer Science and Engineering.
This work was supported in part by the U.S. Army Research Office under Army-ECASE award W911NF-07-R-0003-03, the U.S. Department of Energy, Office of Science, IARPA HAYSTAC Program, NSF Grants #2205093, #2146343, and #2134274, CDC-RFA-FT-23-0069, as well as DARPA AIE FoundSci and DARPA YFA.
Zephyrus: An Agentic Framework for Weather Science , ICLR 2026
Marshall Fisher, Jas Thakker, Yiwei Chen, Zhirui Xia, Yasaman Jafari, Ruijia Niu, Manas Jain, Veeramakali Vignesh Manivannan, Zachary Novack, Luyu Han, Srikar Eranky, Salva Rühling Cachay, Taylor Berg-Kirkpatrick and Rose Yu, Department of Computer Science and Engineering, UC San Diego Jacobs School of Engineering
Duncan Watson-Parris, UC San Diego Scripps Institution of Oceanography and Halicioglu Data Science Institute, within the School of Computing, Information and Data Sciences at UC San Diego
Sumanth Varambally and Yi-An Ma, Halicioglu Data Science Institute within the School of Computing, Information and Data Sciences at UC San Diego
END
New AI agent could transform how scientists study weather and climate
2026-03-10
ELSE PRESS RELEASES FROM THIS DATE:
New study sheds light on protein landscape crucial for plant life
2026-03-10
PULLMAN, Wash. — Research led by scientists at Washington State University has revealed insights on how plants form a microscopic landscape of proteins crucial to photosynthesis, the basis of Earth's food and energy chain.
The discovery provides a new view of the molecular engine that converts sunlight into bioenergy and could enable future fine-tuning of crops for higher yields and other useful traits.
Colleagues at WSU, the University of Texas at Austin, and the Weizmann Institute of Science in Israel used a novel, technology-powered approach to peer inside plant leaf cells and visualize the landscape of the photosynthetic membrane — the ribbon-like structure where plants ...
New study finds deep ocean microbes already prepared to tackle climate change
2026-03-10
CHAMPAIGN, Ill. — Deep-sea waters are warming due to heat waves and climate change, and it could spell trouble for the oceans’ delicate chemical and biological balance. A new study, however, demonstrates that the microbe Nitrosopumilus maritimus may already be adapting well to warmer, nutrient-poor waters. Researchers predict that these surprisingly adaptable iron-dependent ammonia-oxidizing archaea will play an important role in reshaping ocean-nutrient distribution in a changing climate.
The study’s findings are published in the Proceedings of the National Academy of Sciences.
Nitrosopumilus ...
ARLIS partners with industry leaders to improve safety of quantum computers
2026-03-10
COLLEGE PARK, Md. – The Maryland Institute for Quantum Applications (MIQA) at the University of Maryland’s Applied Research Laboratory for Intelligence and Security (ARLIS) has launched a new research initiative under the SEQCURE program, sponsored by the Secretary of the Air Force’s Concepts, Development, and Management Office, to apply Zero Trust Architecture principles to quantum systems.
Working with its industry partners, ARLIS researchers are evaluating the security posture of different environments, providing recommendations to align emerging quantum technologies with national security standards. The six key areas that define the design and use of these ...
Modernization can increase differences between cultures
2026-03-10
Does modernization—economic growth, technological advancement, globalization, increased education, and urbanization—reduce cultural differences? Conventional wisdom suggests that as nations get richer and more educated, a globalized, modern culture emerges featuring low birth rates, high divorce rates, and an overall focus on the individual. Thomas Talhelm tests this hypothesis using the World Values Survey, which has collected data in a broad range of countries since 1981. Notably, variation in values between countries in the World Values Survey has grown from 1981–2017. ...
Cannabis intoxication disrupts many types of memory
2026-03-10
PULLMAN, Wash. — Smoking cannabis can do more than blur memories. It can reshape them.
A new Washington State University study found that people who consumed THC were more likely to recall words that were never presented and struggled with everyday tasks such as remembering to do something later.
Published in the Journal of Psychopharmacology, the study is one of the most comprehensive looks yet at how cannabis affects memory. The findings suggest cannabis can impair not only simple recall, such as remembering a list of words, but also forms of memory people rely on in daily life, like remembering appointments, keeping ...
Heat does not reduce prosociality
2026-03-10
High temperatures have long been empirically linked to violence, conflict, and aggression at the societal level—a troubling pattern in a warming world. Alessandra Cassar and colleagues sought to explore the effect of high heat on individual egalitarianism, resource maximization, selfishness, spite, and competitiveness. The authors invited university students in Colombia, India, Kenya, Mexico, and the United States to play games that involved making choices about whether to share, whether to reduce another player’s payoff at a cost to oneself, as well as whether or not to compete. ...
Advancing brain–computer interfaces for rehabilitation and assistive technologies
2026-03-10
Motor imagery (MI) is the mental process of imagining a specific limb movement, such as raising a hand or walking, without physically performing it. These imagined movements generate distinct patterns of brain activity that can be recorded using electroencephalography (EEG). By decoding these signals, researchers can enable direct communication between the brain and computers, making MI-EEG a powerful tool for applications such as motor rehabilitation and the assistive control of wheelchairs and prosthetic devices.
However, EEG signals generated during MI vary significantly across individuals ...
Detecting Alzheimer's with DNA aptamers—new tool for an easy blood test
2026-03-10
With aging populations on the rise, the need for better tools to diagnose and monitor Alzheimer's disease (AD), the most common cause of dementia, has never been more urgent. This disease is characterized by the gradual loss of nerve cells, a process known as neurodegeneration, which begins years before the onset of obvious symptoms. One way to detect this damage is to look for signs of injury to nerve cells. A key emerging biomarker of neurodegeneration is neurofilament light chain (NfL), a structural protein component ...
Chinese Neurosurgical Journal study develops radiomics model to predict secondary decompressive craniectomy
2026-03-10
Traumatic brain injury (TBI) can trigger dangerous swelling inside the skull, raising intracranial pressure to life-threatening levels even after emergency surgery. In some cases, patients who initially undergo craniotomy to remove a hematoma deteriorate later and require a second, more invasive operation called decompressive craniectomy (DC). Because this escalation often occurs suddenly, clinicians have limited time to respond. Identifying high-risk patients earlier could transform intensive care management, enabling proactive monitoring and timely intervention before irreversible damage occurs.
To address ...
New molecular switch that boosts tooth regeneration discovered
2026-03-10
Dental pulp injury caused by trauma or deep caries often leads to inflammation, tissue necrosis, and eventual loss of tooth vitality. In severe cases, bacterial invasion and sustained immune responses further compromise the pulp’s microenvironment, disrupting its natural capacity for repair. Although regenerative endodontic approaches aim to restore living tissue, predictable biological repair remains difficult to achieve. Central to successful regeneration is the precise regulation of stem cell signaling pathways that coordinate ...